Late marriage and population restraint

Why did some pre-industrial populations stop growing well before their land could no longer feed them?

The standard picture of a pre-industrial population is grim and simple: people breed up to the limit of the food, then starvation, disease and war cut them back. Growth stops because people die.

In parts of north-western Europe in the early modern centuries, that is not what the parish registers show. Populations levelled off, sometimes declined, in periods with no famine, and did so with a signature owing nothing to death rates: women married late, in their mid-twenties rather than soon after puberty, and a substantial share never married at all. So the brake was applied upstream of mortality. Where, and by what?

Ask what a marriage required. Not affection, and not permission in the abstract, but a place: a farm, a tenancy, a workshop, a cottage with a plot. Here a newly married couple were expected to set up their own household rather than move in with his parents — neolocal residence, in the jargon — so marriage waited on their commanding an independent living.

In a village the number of such livings is finite and changes slowly. Farms are fixed, tenancies fall vacant when someone dies or retires, craft positions open when a master needs a successor. A young couple is queuing, and the queue moves at the rate niches open.

Follow the consequence. Delay a woman's first marriage from twenty to twenty-six and six years are gone from the front of her reproductive life — the years fertility is highest and widowhood least likely. Do that across a population, add a substantial fraction who never marry, and the birth rate falls hard without anyone deciding to have fewer children within marriage. That is the crucial point: fertility inside marriage in these populations appears largely uncontrolled. The lever was not contraception but entry into marriage.

Now close the loop, which is where it becomes a system rather than a fact. Suppose harvests are good and real wages rise. A servant saves enough to set up sooner; a landlord subdivides; a craft supports another hand. Niches open, couples marry earlier and more marry at all, births rise. Two decades on that larger cohort reaches the labour market, wages fall back, niches tighten, marriage is deferred, births fall. The population has regulated itself without a single famine.

John Hajnal identified the pattern in 1965 — late female marriage and high permanent celibacy west of a line drawn roughly from Trieste to St Petersburg. Malthus had already named the mechanism: this is the preventive check, restraint acting on births, as against the positive check of mortality, and he thought it the more humane of the two.

Two features make that loop a mechanism rather than a story. It has a long lag — babies born in a good decade do not depress wages until grown, so the system oscillates rather than settles. And Wrigley and Schofield's reconstruction of English population from parish registers found nuptiality doing much of this work, the lag between economic conditions and demographic feedback running to decades. I recall the shape of that finding rather than its figures, and the lag is among the exercise's more contested outputs. This is the mirror of the collection's post-plague wages: there a mortality shock raised wages, here wages govern births.

The evidence needs stating plainly. Almost all of it rests on parish registers of baptisms, marriages and burials, reconstituted into families. Those registers exist only where a church kept them and they survived, so the record is thick in England, France and parts of Scandinavia and thin or absent elsewhere. Reconstitution can only follow people who stayed in one parish long enough to be observed at both ends, which excludes the mobile — disproportionately the poor and the unmarried servants at the heart of this very mechanism. Mean marriage ages quoted for early modern England sit in the mid-twenties for women, but published values move with the parish, the period and the reconstitution rules, so I will not give one as if it were settled.

Three refinements, connecting the steps.

The pattern is regional, not a universal stage of development. In much of eastern and southern Europe, and across most of Asia, women married early and near-universally, and households formed by adding a couple to an existing one. Those populations had other checks; the account explains a place, not a phase of history.

Nor was the restraint deliberate. Nobody needed to intend population control: each couple waited for a living, and the aggregate emerged from millions of separate decisions about a farm or a cottage, which is why the loop works without participants believing in it.

Nor is it the only candidate for the levelling off. Emigration removed people, disease removed more, and land-tenure institutions varied enough to produce similar curves. What makes nuptiality more than coincidence is that it moves in step with wages and ahead of the population change it explains.

What would refute it? The account predicts that age at first marriage and the proportion never marrying respond to real wages and to the availability of holdings, with a lag; that the response appears in nuptiality before marital fertility; and that the pattern weakens where a couple can marry without a niche. So the refuting observations are clear: marriage ages flat through large sustained swings in real wages, or a fertility decline showing up as fewer births per marriage while marriage ages hold steady. The second is roughly what happens in the nineteenth century — not a refutation but a change of regime, and the two are hard to tell apart in aggregate data.

• Why the same region's fertility decline in the nineteenth century worked through births within marriage instead, and what changed to allow it.
• How joint-household rules elsewhere produced different checks on the same pressure.

Every term the collection defines is gathered in the glossary.